diffu/data/__init__.py

"""Data pipeline: (PAGE/ALTO XML  OR  pre-cropped lines) -> (image, text, writer_id) manifest + split."""
from .dataset import (
    HandwritingLineDataset,
    build_manifest,
    char_coverage,
    load_manifest,
    normalize_text,
    write_splits,
    writer_disjoint_split,
)
from .ingest import read_records, scan_folder, writer_from_path
from .parse import LineRecord, parse_alto_xml, parse_file, parse_page_xml

__all__ = [
    "LineRecord",
    "parse_file",
    "parse_page_xml",
    "parse_alto_xml",
    "build_manifest",
    "load_manifest",
    "char_coverage",
    "writer_disjoint_split",
    "write_splits",
    "normalize_text",
    "HandwritingLineDataset",
    "scan_folder",
    "read_records",
    "writer_from_path",
]